First AI Winter
The 1973 Lighthill Report savaged AI research in Britain, and DARPA slashed funding in the United States after promised breakthroughs failed to materialize. Early neural network approaches hit hard theoretical limits exposed by Minsky and Papert’s critique of the perceptron. Researchers who remained in the field quietly shifted toward more modest, narrowly focused problems.
Terry Winograd's SHRDLU could hold a conversation about colored blocks — and fooled everyone into thinking language was solved.
A devastating government review by mathematician James Lighthill declared most AI research a failure, triggering funding cuts that plunged the field into its first winter.
Joseph Weizenbaum published Computer Power and Human Reason, warning that his own ELIZA experiment revealed a dangerous human willingness to trust machines with intimate decisions.
A Stanford AI diagnosed blood infections more accurately than most physicians — then was quietly shelved because no one knew who to blame if it was wrong.
A wobbly Stanford robot took five hours to navigate a chair-filled room, becoming one of the first machines to see and move autonomously.
Philosopher John Searle argued that no computer could ever truly understand language, igniting a debate that still rages today.